Google’s Vertex AI platform has received two new artificial intelligence models – Imagen 3 and Veo. The first one generates images based on a text description, and the second one animates images using a text description. They can work independently or together.
Imagen 3 surpasses previous versions in such aspects as detail, lighting, and artifact reduction. It will be available to all Vertex AI users next week. Below are examples of images created with this model along with the queries used.
Additional features are available for users on the authorized list (you can apply here). They include:
Business users can use reference images along with descriptions. According to Google, the Imagen 3 Customization feature allows you to «integrate your brand, style, logo, subject, or product features when creating new images».
Veo, which is currently available in private preview, lets you animate static images by adding «directorial instructions to them». In addition, Veo can create videos from scratch using only a text query.
The results of both Imagen and Veo models are discreetly watermarked with the DeepMind SynthID watermark, which prevents AI images from being passed off as real. Security filters ensure that AI does not create malicious content. Google also notes that no customer data was used to train its AI, and the company provides protection against copyright claims.
Additional information is available at by the following link.
Source: gsmarena